Chinese AI tools now rival Anthropic Mythos in vulnerability detection as gap narrows

Chinese AI Closes Gap with Anthropic Mythos in Vulnerability Discovery

Chinese AI has reached a pivotal milestone in the US-China AI race, with multiple systems now matching Anthropic Mythos in detecting software security vulnerabilities. At the ISC.AI 2026 cybersecurity conference in Beijing on June 24, 360 Security Technology founder Zhou Hongyi unveiled two AI-driven cybersecurity tools collectively named "Yitian Tulong"—a reference to a classic Chinese martial arts novel meaning "Heavenly Sword and Dragon Saber." 1

1

The first tool, Tulongfeng, focuses on automated vulnerability discovery and was explicitly positioned as "China's version of Mythos," while the second, Yitianzhen, handles automated cyber defense and incident response. 2

2

Zhou Hongyi framed vulnerability-finding AI as a national strategic asset, declaring that "this kind of powerful weapon that can change the landscape of cyber offence and defence cannot be held only by others." 1

1

His remarks reflect widespread unease in China about what state media has called the "unprecedented cyberattack capabilities" displayed by Anthropic Mythos, which was previewed in April and reportedly found thousands of major vulnerabilities in operating systems, web browsers, and other software. 4

4

360 Security Technology Claims 3,432 Vulnerabilities Found

360 Security Technology claims Tulongfeng has discovered 3,432 software vulnerabilities, including 105 confirmed by Chinese authorities, though Reuters could not independently verify these claims. 1

1

Zhou acknowledged that domestic models still have a 20-30% gap in base capability compared to American rivals, but insisted China could not wait for full parity before pursuing vulnerability discovery. 2

2

Instead, 360 is taking an "agent" route, combining models with security expertise, vulnerability databases, and automated tools—an approach Zhou claimed gives Tulongfeng "Mythos-equivalent capabilities." 4

4

"If Mythos is a top-end chip, what we are building is a complete machine that can run stably, work 24 hours a day and make fewer mistakes," Zhou explained. "If the U.S. route is to cultivate a genius hacker, 360's route is to organise a professional attack-and-defence team." 1

1

Zhipu AI's GLM-5.2 Matches Mythos in Bug Detection

Separately, security researchers told The Wall Street Journal that Zhipu AI's GLM-5.2 model now performs on par with Anthropic Mythos in identifying software bugs, even as it trails in broader reasoning tasks. 3

3

GLM-5.2 is an open-weight model, meaning anyone can download, modify, and run it on their own hardware without relying on cloud providers—a flexibility that appeals to enterprises but also raises national security concerns about potential misuse by cybercriminals. 5

5

Cybersecurity company Semgrep found that GLM-5.2 outperformed Anthropic's Claude Opus 4.8 in some benchmark tests, while researchers discovered that with additional prompting, both Opus 4.8 and GLM-5.2 can match Mythos in bug detection. 5

5

According to OpenRouter, which provides access to more than 400 AI models, GLM-5.2 ranks among the 10 most-used AI systems globally. 5

5

National Security Concerns and Export Controls Fuel Development

The rapid advancement in Chinese AI capabilities comes amid tightening U.S. export controls on China's access to cutting-edge chips since 2022, restrictions the U.S. has justified by arguing these chips would allow the Chinese military to enhance its capabilities with AI. 1

1

This month, the U.S. government ordered Anthropic to suspend exports of a less powerful version of Mythos to destinations worldwide and all foreign nationals due to national security concerns. 4

4

Zhou argued that China faced a risk of "one-way transparency" if U.S. entities could use Mythos-like models to scan software and critical systems while Chinese companies were denied comparable capabilities. 1

1

His concerns are amplified by data showing 67% of 1,000 executives surveyed in an IBM and Palo Alto Networks study said they had been targeted by AI attacks within the past year. 4

4

Geopolitical Tensions Shape AI Development Strategies

China and the U.S. have a long history of accusing each other of conducting offensive cyber operations on critical infrastructure, and 360's release marks the most high-profile Chinese answer yet to Anthropic's Mythos model. 1

1

Lior Div, chief executive of cybersecurity company 7AI, told The Wall Street Journal that "China is making sure that the gap becomes smaller and smaller over time." 5

5

Critics have argued that the administration's approach is counterproductive, particularly as it has allowed exports of AI chips to China despite the country's rapid AI advances. Saif Khan, a distinguished technology fellow at the Institute for Progress who worked on export restrictions during the Biden administration, stated: "Banning Fable while selling chips China needs to develop its own version is a gift to China." 5

5

Meanwhile, a flood of high-powered, cheap-to-use Chinese open-weight models are quickly drawing customers across the U.S., with companies including Microsoft considering integrating these systems on their platforms. 5

5